Meta Interview Questions — Page 124

Meta·Data Scientist·Technical Phone Screen

May 2026

Meta DS interview with a probability and statistics question centered on ad insertion strategies in a news feed. The problem had multiple parts and required solid knowledge of binomial distributions, variance, and tail probabilities. Felt like a stats exam more than a typical DS screen.

You have a 100-post news feed. Option A inserts ads by giving each post an independent 4% chance of being an ad. Option B places exactly one ad per every 25 posts. For both options, calculate the expected number of ads and the variance. Then find the probability that the number of ads exceeds twice the expected count. Finally, estimate the expected number of adjacent ad pairs for each option and compare which is more likely to produce long runs of consecutive ads.

“This is a four-part question dressed up as one, which I did not fully appreciate until I was halfway through the variance calculation and realized I hadn't even thought about the adjacent pairs part yet.”

View Post

Meta·Data Scientist·Onsite - Product Sense / Strategy

May 2026

This was a product-heavy data science interview at Meta that threw me into a full-scale feature build decision for a group call product. One question, but it was basically a whole case study compressed into a single prompt. Walked out unsure if I'd said enough or way too much.

You need to decide whether to build a group call feature without relying on existing product usage tables. What non-table data sources would you pull from, what signal would you extract from each, and how would you quantify and de-bias them? Then design a full experiment plan to validate demand, including a fake-door test, a waitlist, and a limited beta, with hypotheses, eligibility criteria, success metrics, guardrails, sample size math, and a timeline anchored to today. Also cover how you'd segment results, control for novelty effects, and handle ethics.

“This question is enormous and I did not fully appreciate that until I was about four minutes in and realized I'd only covered support tickets.”

View Post

Meta·Software Engineer·Technical Phone Screen

May 2026

Meta coding round for a software engineer position. One linked list problem that looks clean on the surface but has enough edge cases to trip you up if you're not careful about pointer management.

Given the head of a singly linked list and a value v that appears at least twice, reverse the sublist between the first and second occurrence of v (inclusive), in-place with O(n) time and O(1) space. How do you handle edge cases like adjacent nodes, or when the sublist starts at the head or ends at the tail?

“I started with the standard reverse-sublist approach and felt pretty confident until they asked about the head case.”

View Post

Meta·Software Engineer·Onsite - System Design / Architecture

May 2026

Meta system design interview focused entirely on WhatsApp Status, which sounds straightforward until they ask you to build search on top of ephemeral content. The search portion is where things got interesting and honestly a bit brutal.

Design the WhatsApp Status feature, covering post creation, viewing, 24-hour expiration, view tracking, and privacy controls.
How would you design search across WhatsApp Statuses, supporting queries by text caption, hashtag, and author, while respecting privacy boundaries and handling the ephemeral nature of the content?

“The core feed stuff felt manageable.”

View Post

Meta·Software Engineer·Onsite - Behavioral / Leadership

May 2026

Behavioral round at Meta for a software engineer role. Just one question but they really wanted you to dig into the details of what you actually built.

Tell me about your favorite project. Walk me through the scope and complexity of what you worked on.

“I picked something I was genuinely proud of but kind of fumbled the complexity angle.”

View Post

Meta·Data Scientist·Onsite - Behavioral / Leadership

May 2026

Behavioral round for a Data Science role at Meta. The question was a single sprawling scenario about team conflict and psychological safety, the kind that sounds like an HR training module but with enough edge cases layered in that it genuinely tripped me up.

A teammate privately tells you they feel unwelcome by a subgroup on your team. Walk through what you do in the first 72 hours: how do you assess the situation, protect psychological safety, gather evidence without outing the reporter, what do you say in 1:1s with each side, what metrics or artifacts do you create to track improvement over 60 days, and how do you document your decisions? Also: what if leadership pushes to move the reporter to another team, and what if the accused group denies everything and you suspect retaliation in a remote-first setup?

“There's a lot packed into this and I made the mistake of trying to address every sub-question linearly, which made my answer feel like a checklist rather than a judgment call.”

View Post

Meta·Software Engineer·Onsite - Coding / Algorithms

May 2026

Meta SWE coding round, got a bitmask-style string problem that looked deceptively simple at first glance. The constraint on unique letters across the whole concatenation is the part that trips people up.

Given an array of strings, select a subset such that no letter appears more than once across the entire concatenation. Return the maximum possible total length of such a subset.

“The aa case is what gets you if you're not paying attention.”

View Post

Meta·Software Engineer·Technical Phone Screen

May 2026

Meta SWE coding round with an interactive grid search problem. The twist is you can only probe the grid through a black-box API, so you have to think carefully about how many calls you're burning.

Given an N×N grid hiding a single length-3 battleship (placed horizontally or vertically), find all three occupied cells. You can only interact with the grid through a bomb_location(x, y) API that returns true on a hit and false on a miss. Implement find_battleship(N) and discuss worst-case API call complexity.

“My first instinct was to just scan every cell row by row which is O(N^2) calls and technically correct but they pushed back immediately asking if I could do better.”

View Post

Meta·Data Scientist·Technical Phone Screen

May 2026

Meta data scientist interview with a meaty stats/probability question that combined Bayes' theorem, confusion matrices, and real arithmetic. The kind of question that sounds manageable until you're actually doing the math live.

You have a classifier that flags potentially bad actors on a platform. Given a prior probability p that any user is a bad actor, and known true-positive and false-positive rates for the classifier, derive the posterior probability that a flagged user is actually bad. Then define Type I and Type II errors in this context, explain the business tradeoffs each represents, and work through a concrete example: 1% of 10 million users are bad actors, the classifier has 95% recall and a 2% false-positive rate. How many bad actors get caught, and how many good users get wrongly flagged?

“I started okay with the Bayes derivation but fumbled the arithmetic partway through.”

View Post

Meta·Software Engineer·Onsite - Coding / Algorithms

May 2026

Meta SWE coding round with one algorithmic problem about sorting and minimum moves. Not a lot of context in the original post but the problem itself was interesting enough to think about.

Given an unsorted array like [5, 2, 2, 4, 3], what is the minimum number of move operations needed to transform it into a sorted array, where each operation moves one element from its current position to another?

“My first instinct was to think about it like a sorting algorithm and count swaps, which was wrong.”

View Post